A Lightweight Cache Insertion Filtering Scheme for Information- Centric Networking
Abstract— The success of Information-Centric Networking owes much to the feasibility of large caches. As such, solid state disk (SSD) is widely adopted as a cache device to scale up the cache size up to terabytes in some recent works. Nevertheless, the limited lifetime is a significant drawback of SSD, making it non-trivial to design a cache system involving SSD. Hence, writing operations on SSD should be elaborately considered in order to reduce the frequency of writes without compromising cache hit ratio. This may seem contradictory, but it is possible since the typical Internet traffic patterns follow the Zipf distribution. This characteristic can be fully exploited to store a small set of popular contents. For this reason, this paper proposes a lightweight cache insertion filtering scheme based on least recently used (LRU) queue and hash table. Our goal is to prevent these unpopular contents from entering into the cache and only store high frequently requested contents. The experimental results show that our proposed scheme can reduce the number of writing operations on SSD without comprising cache hit ratio. In fact, it can improve cache hit, compared with the unfiltering scheme and a lightweight probationary insertion filtering scheme, while only requiring a modest memory and consuming about two hundred CPU cycles.
Index Terms— Information-Centric Networking, SSD, filtering, cache hit ratio, CPU cycle.
Department of Automation, University of Science and Technology of China, CHINA
Jinlin Wang, Lingfang Wang, Yiqiang Sheng
National Network New Media Engineering Research Center, Institute of Acoustics, Chinese Academy of Sciences, CHINA
Cite: Li Ding, Jinlin Wang, Lingfang Wang, Yiqiang Sheng, "A Lightweight Cache Insertion Filtering Scheme for Information- Centric Networking," Proceedings of 2018 the 8th International Workshop on Computer Science and Engineering, pp. 453-460, Bangkok, 28-30 June, 2018.